How to Get 200% Out of Gemini: Practical AI Tips from a ChatGPT Comparison

[Just listen!]

Stop_Asking_Which_AI_Is_Best__The_Expert_Strategy_for_Using_Cha

0:00

Moving Beyond "Which AI Should I Use?" to "How Should I Use AI?"

What AI should I use?

As someone who uses and recommends AI daily, the most common question I get is, "Which AI is the best?" While the market is flooded with countless AI tools, for general work or daily life, it ultimately boils down to two giants: OpenAI's ChatGPT and Google's Gemini. This article aims to move beyond the simple hype and share my personal experience to answer a more critical question: "How do I choose the right tool for my work and use it like an expert?"

This article is based on my hands-on experience testing and using both models in various real-world scenarios, from data crawling and image generation to work management.

The First Conversation with AI: Intuitive Understanding vs. Massive Information Processing

ChatGPT: The Best Partner for Beginners

ChatGPT: The Best Partner for Beginners

For those just starting with AI, I recommend ChatGPT without hesitation. The reason is simple: it "just gets you." This intuitive and human-like conversational ability is the biggest appeal of OpenAI's model. Even if you phrase things awkwardly, ChatGPT accurately grasps the meaning and elaborates on it.

When you ask it to write a difficult introduction or give it a jumbled bunch of thoughts, it acts less like a simple machine and more like a creative partner you can always lean on, understanding you perfectly to produce unexpected text or neatly organize your messy ideas. However, this strength also leads to a higher frequency of making things up—the 'Hallucination' phenomenon—often resulting in inaccurate information.

Gemini: The Data Analyst's Hidden Assistant

Gemini, a token with plenty of room

I discovered the true power of Gemini while performing a large-scale data crawling task. In a test to scrape data from 50 marketing agencies, ChatGPT surprisingly had a 60% error rate, while Gemini's was only 40%. (Note: Results can vary depending on the prompt, updates, and other factors.)

A likely reason for this is Gemini's ability to remember a staggering 1 million tokens at once, whereas GPT offers 200,000 tokens. In my crawling job, the AI with the smaller memory capacity would "forget" the initial instructions towards the end of the task, leading to bizarre results. In fact, in various tests with extensive documents, I often found Gemini outperforming GPT-4.

So now, I turn to ChatGPT when I need creative writing and to Gemini when I need to summarize market research reports or analyze data. However, for some specific functions, like Korean Optical Character Recognition (OCR) from images, GPT-4 often proves superior.

Differences in Image Generation: Gemini vs. GPT-4

The difference in image generation

One of the most popular and beloved features of AI is likely image generation. Here is what I've learned through continuous and varied image creation experiences.

Gemini's Specialty: Multiple Images from a Single Command!

My favorite image generation feature in Gemini is its ability to create multiple drafts simultaneously from a single command. For example, if you say, "Create 4 images of a dog surfing," it generates four different images. This allows me to create and use images quickly.

Especially after the recent 'Nano Banana' update, it has been praised for producing top-tier image quality. A great tip is that if you combine two images or have a reference image, anyone can easily create high-quality results.

However, there are downsides. While you can use Korean for all commands and conversations, any text included in the images often comes out garbled or incorrect. When creating multiple images, they all tend to be in a very similar style, limiting the ability to produce a diverse range of artistic styles. A notable issue is that even with the same prompt, it consistently produces different images, making it difficult to achieve consistent quality without specific criteria.

Gemini vs. ChatGPT Table

	Gemini (Imagen 3)	ChatGPT (DALL-E 3 / GPT-Image-1)
Feature	Can generate multiple images at once.	Can only generate one image at a time.
User Interface	Select the image icon to create (Easy).	Use the 'create image' keyword in a prompt to create (Very Easy).
Weakness	Watermark on the right, frequent errors.	Difficulty in progressing due to the need for continuous prompt creation.

The Power of the Ecosystem: Perfect Integration with Google Workspace

The power of the Google Gemini ecosystem

The decisive reason I use Gemini as my primary tool is its integration with Google Workspace. This is Gemini's unique and powerful weapon.

Not Just 'Linked,' but 'Fused': A Revolution in My Workflow

When I think about what consumes the most time at work, it's the 'context switching' cost that comes from hopping between different apps. You open a web browser to find data, switch back to a document to write, then open your email to send it your focus is constantly broken. Gemini frees you from this problem.

With Gemini, you can summarize a Google Docs document without leaving your Gmail window, use that summary to draft a reply, and pull relevant data directly from Sheets. These steps might seem minor, but they dramatically increase work focus and efficiency.

Tools You Can Use with AI: Gmail & Google Sheets & Google Slides

1. Gmail & Gemini

• Email Summary: When a long, complex English email thread arrives, I can grasp the key points with a single click.

• AI Draft ("Help me write"): Just by throwing in a few key points, Gemini whips up a formal email draft. For example, "Politely write a follow-up email for the Q3 proposal based on our call last Tuesday."

• Context-Aware Smart Reply: It goes beyond the old single-word suggestions, understanding the full context of the email to propose detailed draft replies.

• The Magic Between Apps: The most powerful feature is pulling information directly from Drive. Prompts like, "Draft a reply to this email and include the key points from the Google Docs document @'Q3 Marketing Report'" can maximize work efficiency.

2. Google Sheets: Data is No Longer Intimidating

• Instant Tables and Planners: Create project trackers, event schedules, or budget plans with a simple sentence like, "Create a July social media content calendar with columns for Platform, Content Type, Post Date, and Status."

• Natural Language Formulas: I used to struggle with complex functions like =VLOOKUP, but now I can just ask in plain language, "Create a formula that looks for the value in cell C1 within the range D:G and returns the corresponding value from column G."

• Quick Analysis and Charts: You can ask questions directly about your data. Get instant visualizations with prompts like, "Analyze my expenses for this month and create a pie chart showing the percentage of each category."

3. Google Slides: Visualizing Ideas

• From Document to Presentation: A single prompt like, "Create a 5-slide presentation based on the document @'Phoenix Project Proposal'" can generate an entire presentation from an existing Google Docs file. However, it still has some weaknesses in terms of fine details.

• On-the-fly Image Generation: Create custom images that match your presentation's theme without ever leaving Slides. You can be specific: "Generate an image of a diverse team collaborating in a modern office using our brand colors #4285F4 and #34A853."

• Summaries and Speaker Notes: Automatically generate a concise summary of your presentation or create detailed speaker notes for each slide.

From Error to Expert: My Advanced Google AI Toolkit

Finding the Hidden Tools

The reason I didn't give up on Gemini, despite the initial frustration and frequent errors, was because I discovered that Google has much more powerful tools hidden away for experts.

1. Google AI Studio: My Personal AI Laboratory

AI Studio is a space designed to let you select the exact AI capabilities you want, allowing you to harness 100% of a model's performance. It's the perfect place to use AI's functions most effectively and without errors.

Why I use AI Studio:

• Direct Model Selection: I can test the same prompt with different models like Gemini 1.5 Pro and 1.5 Flash to find the optimal result.

• Personality Control: By adjusting the 'temperature' value, I can make the AI's responses more creative or, conversely, more consistent.

• Custom Chatbot Creation: I can input sophisticated system instructions to pre-build and test my own purpose-specific chatbots without any coding.

• Convert to Code: Once my prompt testing is done, I just click the 'Get Code' button to get API code that I can immediately paste into my app.

• Exploring Advanced Features: I can use screen sharing to work on tasks together with the AI, as it sees what I see.

2. NotebookLM: A Hallucination-Free AI, My Personal Knowledge Expert

If AI Studio is my lab, NotebookLM is my personal library. This tool fundamentally solves AI's biggest headache—'hallucination' (making things up)—by ensuring it only finds information within the sources you provide. When you upload PDFs, Google Docs, or website links into a specific 'notebook,' the AI answers questions based only on that material.

Why I love NotebookLM:

• Verifiable Answers: Every answer NotebookLM provides comes with a clear citation. With a single click, I can check the original source text, making its reliability 100%.

• Information Fusion: I can upload several market reports at once and ask, "Compare and analyze the commonalities and differences between these reports," and get integrated insights in seconds.

• Automatic Content Generation: It automatically creates FAQs, study guides, or briefing documents based on the materials I've uploaded, turning simple information into structured knowledge.

• Multimedia Understanding: Now, you can even ask questions and get answers about the images, charts, and diagrams within your documents.
  

NotebookLM is the 'secret weapon' I strongly recommend for anyone who needs to become an expert in a specific field quickly and accurately, such as students analyzing research papers, lawyers reviewing case law, or consultants synthesizing client data.

It can also create video and audio materials to help you understand your original source material, much like the audio guide (broadcast) mentioned at the very top of this article. Additionally, it offers various features like quizzes and mind maps to help you comprehend information more easily.

My Personal Know-How for Dealing with "Constant Errors"

AI's persistent errors

When an AI continuously generates errors, you can troubleshoot them using the following methods:

• Review the Prompt: Errors are more likely when you ask for multiple tasks at once. Try to request only one task at a time.

• Prepare the Prompt: Since the same prompt can yield different results, use copy-and-paste to reuse an identical prompt multiple times.

• Acknowledge the Problem: If the AI's answer is poor, point out its mistake and use negative feedback like, "Is that the best you can do?"

• The Art of the Break: If you encounter persistent errors, try using a phrase that tells it to rest before answering, like "Take a break and then try again."

• Add a Review Step: Include a phrase that asks the AI to double-check if its response meets the request, forcing it to review its own answer.

• The Most Reliable Method: Since AI builds upon previous results, the most effective solution is often to simply open a new chat window.

The Art of Choosing the Right AI for Your Needs

A balanced AI choice

Now, when someone asks me whether Gemini or ChatGPT is better, I ask them back, "What kind of task are you doing?" The criteria for choosing a tool ultimately depend on the nature of the work.

Based on my experience, here is my final advice:

• Use ChatGPT when: You need creative ideas and want to develop your thoughts through open-ended conversations. I recommend using it primarily as a brainstorming partner or for drafting initial content.
  

• Use Gemini when: You are doing analytical, data-centric work or want to maximize your daily work efficiency. I suggest getting help from Gemini when researching extensive documents, quickly analyzing data with Sheets, and automating communication in Gmail and Slides.

I believe the true experts of the AI era won't be tied to a single brand, just like a skilled craftsman knows the unique characteristics of every tool in their toolbox. I hope this article serves as a small starting point for your AI journey. Don't stop here; keep experimenting and pushing the boundaries. I'm rooting for you to become someone who doesn't just use AI, but strategically deploys the optimal AI for every situation.